Inspiration
We were inspired to do this project because of the rise of machine learning and algorithmic writing. Algorithmic writing has the potential to displace jobs, but what if people worked together with algorithmic writing?
Using our project, we were able to incorporate the efforts of both the late writer Edward Estlin "E. E." Cummings with that of the writings of the user.
This project is meant to demonstrate that it is possible for a writer to collaborate and further extend their writing skills with the assistance of machine learning.
What it does
Our project is a website that:
- displays recently submitted user submissions of machine-learning-assisted poetry.
- allows users to generate poetry from e.e. cummings and submit their modified poetry.
How we built it
We used a github implementation that would take text as an input, which then outputted something that was statistically similar. This implementation used Recurrent Neural Network learning, so it started with one word and then generated words after each other based off word probability. Using enough training, it was able to formulate somewhat coherent phrases.
We used this algorithm to generate a 20,000 word text database of poems that our website would easily access. Our website was able to generate text based off this database for the user. Once the user modified their poem, they were able to create a new poem inspired by the generated poem. This poem could be published, and this went into our firestore database. On the main page, we were able to access our database and we displayed this data on our main page.
Challenges we ran into
We don't like writing poetry
We came in not knowing a single thing about html, js, css, or databases. We had to ask just about every mentor that we could find to make all this work out. Thanks to all the mentors who came to LA hacks! You made our lives so much easier.
Domains.com did not work. I used a private network to access it, and I got a free domain. However, that did not work, as they did not let me upload my files.
Accomplishments that we're proud of
We're really proud that we made this thing work at all. However, our biggest pride was in making the database work, as this was a daunting task and the data storing and retrieving meant that we had to learn a lot of new things.
What we learned
We learned how to use HTML, JS, CSS, and firestore. My teammate and I were used to making Unity projects, so this was definitely outside of our comfort zone.
What's next for Poet Tree
We're planning on changing the look so that it looks better. We are also thinking about adding other poets, and getting people to use this.
Built With
- css
- firestore
- html5
- javascript

Log in or sign up for Devpost to join the conversation.